mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: "H. Peter Anvin" <hpa@zytor.com>
To: John Denker <jsd@av8n.com>,
	tytso@mit.edu, noloader@gmail.com, linux-kernel@vger.kernel.org,
	Stephan Mueller <smueller@chronox.de>,
	Herbert Xu <herbert@gondor.apana.org.au>,
	andi@firstfloor.org, Sandy Harris <sandyinchina@gmail.com>,
	cryptography@lakedaemon.net, linux-crypto@vger.kernel.org
Subject: Re: [PATCH 1/3] random: replace non-blocking pool with a Chacha20-based CRNG
Date: Wed, 04 May 2016 14:56:14 -0700	[thread overview]
Message-ID: <C6DC09E0-98EA-4591-9336-4D2BE1699C2C@zytor.com> (raw)
In-Reply-To: <572A6CDD.10503@av8n.com>

On May 4, 2016 2:42:53 PM PDT, John Denker <jsd@av8n.com> wrote:
>On 05/04/2016 12:07 PM, tytso@thunk.org wrote:
>
>> it doesn't hit the
>> UB case which Jeffrey was concerned about. 
>
>That should be good enough for present purposes....
>
>However, in the interests of long-term maintainability, I
>would suggest sticking in a comment or assertion saying 
>that ror32(,shift) is never called with shift=0.  This
>can be removed if/when bitops.h is upgraded.
>
>There is a track record of compilers doing Bad Things in
>response to UB code, including some very counterintuitive
>Bad Things.
>
>On Wed, May 04, 2016 at 11:29:57AM -0700, H. Peter Anvin wrote:
>>>
>>> If bitops.h doesn't do the right thing, we need to
>>> fix bitops.h.
>
>Most of the ror and rol functions in linux/bitops.h
>should be considered unsafe, as currently implemented.
>http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/include/linux/bitops.h?id=04974df8049fc4240d22759a91e035082ccd18b4#n103
>
>I don't see anything in the file that suggests any limits
>on the range of the second argument.  So at best it is an
>undocumented trap for the unwary.  This has demonstrably
>been a problem in the past.  The explanation in the attached
>fix-rol32.diff makes amusing reading.
>
>Of the eight functions
>  ror64, rol64, ror32, rol32, ror16, rol16, ror8, and rol8,
>only one of them can handle shifting by zero, namely rol32.
>It was upgraded on Thu Dec 3 22:04:01 2015; see the attached
>fix-rol32.diff.
>
>I find it very odd that the other seven functions were not
>upgraded.  I suggest the attached fix-others.diff would make
>things more consistent.
>
>Beware that shifting by an amount >= the number of bits in the
>word remains Undefined Behavior.  This should be either documented
>or fixed.  It could be fixed easily enough.

This construct has been supported as a rotate since at least gcc2.
-- 
Sent from my Android device with K-9 Mail. Please excuse brevity and formatting.

  parent reply	other threads:[~2016-05-04 21:56 UTC|newest]

Thread overview: 53+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-05-02  6:26 [RFC PATCH 0/3] random: replace urandom pool with a CRNG Theodore Ts'o
2016-05-02  6:26 ` [PATCH 1/3] random: replace non-blocking pool with a Chacha20-based CRNG Theodore Ts'o
2016-05-03  8:50   ` Stephan Mueller
2016-05-04 16:54     ` Jeffrey Walton
2016-05-04 17:30     ` tytso
2016-05-04 17:52       ` H. Peter Anvin
2016-05-03  9:36   ` Stephan Mueller
2016-05-04  6:24     ` Stephan Mueller
2016-05-04 14:40   ` Jeffrey Walton
2016-05-04 17:49     ` tytso
2016-05-04 18:22       ` Jeffrey Walton
2016-05-04 18:29         ` H. Peter Anvin
2016-05-04 19:07           ` tytso
2016-05-04 20:53             ` H. Peter Anvin
2016-05-04 21:42             ` John Denker
2016-05-04 21:52               ` better patch for linux/bitops.h John Denker
2016-05-05  1:35                 ` Jeffrey Walton
2016-05-05  2:41                   ` H. Peter Anvin
2016-05-05  2:54                     ` Jeffrey Walton
2016-05-05  3:08                       ` H. Peter Anvin
2016-05-05  3:30                         ` Jeffrey Walton
2016-05-05  3:50                           ` Theodore Ts'o
2016-05-05  4:03                             ` Jeffrey Walton
2016-05-05  6:35                               ` H. Peter Anvin
2016-05-05 16:15                                 ` UB in general ... and linux/bitops.h in particular John Denker
2016-05-05 17:32                                   ` Andi Kleen
2016-05-06  2:25                                   ` Jeffrey Walton
2016-05-05 21:34                             ` better patch for linux/bitops.h Sandy Harris
2016-05-05 22:18                               ` tytso
2016-05-05 22:22                                 ` H. Peter Anvin
2016-05-05 22:38                                 ` H. Peter Anvin
2016-05-06  0:13                                 ` H. Peter Anvin
2016-05-04 21:56               ` H. Peter Anvin [this message]
2016-05-04 22:06                 ` linux/bitops.h John Denker
2016-05-04 23:06                   ` linux/bitops.h Andi Kleen
2016-05-05  0:13                     ` linux/bitops.h John Denker
2016-05-05  1:20                     ` linux/bitops.h Jeffrey Walton
2016-05-05  1:27                       ` linux/bitops.h H. Peter Anvin
2016-05-05  0:30                   ` linux/bitops.h H. Peter Anvin
2016-05-05  0:48                     ` linux/bitops.h Linus Torvalds
2016-05-06 20:08                       ` linux/bitops.h Sasha Levin
2016-05-06 20:07                     ` linux/bitops.h Sasha Levin
2016-05-06 20:25                       ` linux/bitops.h H. Peter Anvin
2016-05-06 20:30                       ` linux/bitops.h H. Peter Anvin
2016-05-02  6:26 ` [PATCH 2/3] random: make /dev/urandom scalable for silly userspace programs Theodore Ts'o
2016-05-02  7:00   ` Stephan Mueller
2016-05-02 12:50     ` Theodore Ts'o
2016-05-02 13:48       ` Theodore Ts'o
2016-05-02 13:53         ` Stephan Mueller
2016-05-02  6:26 ` [PATCH 3/3] random: add interrupt callback to VMBus IRQ handler Theodore Ts'o
2016-05-02  9:00   ` Jeffrey Walton
2016-05-02  9:14     ` Stephan Mueller
2016-05-02 12:56       ` Theodore Ts'o

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=C6DC09E0-98EA-4591-9336-4D2BE1699C2C@zytor.com \
    --to=hpa@zytor.com \
    --cc=andi@firstfloor.org \
    --cc=cryptography@lakedaemon.net \
    --cc=herbert@gondor.apana.org.au \
    --cc=jsd@av8n.com \
    --cc=linux-crypto@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=noloader@gmail.com \
    --cc=sandyinchina@gmail.com \
    --cc=smueller@chronox.de \
    --cc=tytso@mit.edu \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®